The emergence of modern sea ice cover in the Arctic Ocean

Received 1 Jul 2014

|

Accepted 20 Oct 2014

|

Published 28 Nov 2014

The emergence of modern sea ice cover

in the Arctic Ocean

Jochen Knies

1,2

, Patricia Cabedo-Sanz

3

, Simon T. Belt

3

, Soma Baranwal

2

,

Susanne Fietz

4,5

& Antoni Rosell-Mele

´

4,6

Arctic sea ice coverage is shrinking in response to global climate change and summer ice-free

conditions in the Arctic Ocean are predicted by the end of the century. The validity of this

prediction could potentially be tested through the reconstruction of the climate of the

Pliocene epoch (5.33–2.58 million years ago), an analogue of a future warmer Earth. Here we

show that, in the Eurasian sector of the Arctic Ocean, ice-free conditions prevailed in the early

Pliocene until sea ice expanded from the central Arctic Ocean for the first time ca. 4 million

years ago. Amplified by a rise in topography in several regions of the Arctic and enhanced

freshening of the Arctic Ocean, sea ice expanded progressively in response to positive

ice-albedo feedback mechanisms. Sea ice reached its modern winter maximum extension for

the first time during the culmination of the Northern Hemisphere glaciation, ca. 2.6 million

years ago.

DOI: 10.1038/ncomms6608

1_{Department of Marine Geology, Geological Survey of Norway, NO-7491 Trondheim, Norway.}2_{Centre for Arctic Gas Hydrate, Environment and Climate}

(CAGE), Department of Geology, University of Tromsø, NO-9037 Tromsø, Norway.3Biogeochemistry Research Centre, School of Geography, Earth and Environmental Sciences, University of Plymouth, Plymouth PL4 8AA, UK.4_{Institut de Cie`ncia i Tecnologia Ambientals, Universitat Auto`noma de Barcelona,}

08193 Bellaterra, Catalonia, Spain.5_{Department of Earth Sciences, Stellenbosch University, Stellenbosch 7600, South Africa.}6_{Institucio´ Catalana de Recerca}

i Estudis Avanc¸ats (ICREA), 08010 Barcelona, Catalonia, Spain. Correspondence and requests for materials should be addressed to J.K. (email: jochen.knies@ngu.no).

Q

uantitative determination of past sea ice coverage in the

Arctic Ocean is critically important for future climate

predictions, as sea ice constrains the effect of changing

surface albedo, ocean–atmosphere heat exchanges and potential

freshwater export to the North Atlantic which, in turn, influences

the meridional overturning circulation (MOC) and thus, global

climate

1

. Through the Integrated Ocean Drilling Program Arctic

Coring Expedition

2

, the prevalence of sea ice in the central Arctic

Ocean since the middle Miocene (B18 Ma) has been established

previously

3–5

, although new research proposes the ephemeral

presence of perennial sea ice as early as 36 Ma (ref. 6).

A reduction in sea ice cover has been reported during the

upper Miocene

7

and the Pliocene

8

, and modelling experiments

indicate that the removal of Arctic sea ice around the early-to-late

Pliocene transition (B3.6 Ma) explains the proxy-based high

paleotemperature reconstruction in the Canadian high Arctic

9,10

.

Other simulations, however, reveal reduced, but still significant

Arctic sea ice cover during the same period

11

. Dowsett et al.

12

concluded that the present discrepancy between Pliocene sea

surface temperature (SST) proxy estimates and model simulations

in the high latitudes requires a further refinement of the current

model parameterization. Since Pliocene sequences in the Arctic

are incomplete, the spatial and temporal evolution of SST and sea

ice extent remains, however, largely uncertain.

In the current study, we investigated long-term changes in

Arctic sea ice coverage at its present summer ice margin in the

Atlantic—Arctic gateway (AAG) region using borehole data that

expose a complete marine Pliocene sequence. Ocean Drilling

Program Hole 910C (80°15.896

0

_{N, 6°35.430}

0

_{E, 556 m water}

depth) and Hole 911 A (80°28.466

0

N, 8°13.640

0

E, 902 m water

depth) were recovered from the eastern flank of the Yermak

Plateau, NW Spitsbergen (Fig. 1)

13

. This region is particularly

well suited for the study of sea ice limits in the Arctic Ocean

because it is located within the spatial and temporal dynamics of

the inflow of warm Atlantic water and outflow of cold Arctic

water masses via the Transpolar Drift (Fig. 1). In recent years, the

analysis of the biomarker IP

25

(‘Ice Proxy with 25 carbon

atoms’)

14

, a mono-unsaturated highly branched isoprenoid lipid

biosynthesized by certain sea ice diatoms

15

, has become an

established proxy method for the reconstruction of Arctic sea

ice

16

. The presence (absence) of IP

25

in Arctic marine sediments

is especially sensitive to the past occurrence (absence) of sea ice

16

,

while

abundance

variations

are

usually

consistent

with

corresponding directional changes in sea ice cover

16

. These

attributes are of particular significance to the current study, as it

has previously been shown that ice marginal conditions in the

gateway (Fram Strait) are mirrored, qualitatively, by the

occurrence of IP

25

in marine sediments from the study region,

and that sedimentary abundances of IP

25

are sensitive to changes

in sea ice conditions

17,18

, thus permitting semi-quantitative

reconstructions. As a result, the measurement of IP

25

in AAG

sediments covering the last 2 Ma has been used to imply marginal

sea ice conditions during the Pleistocene

19

. The new IP

25

record

from the AAG shows that sea ice expanded from the Arctic

Ocean to its modern limits for the first time ca. 4 million years

ago. We suggest that a rise in topography above a critical

threshold for ice accumulation in several regions of the Arctic,

together with enhanced freshwater delivery to the Arctic

Ocean during the early Pliocene preconditioned the northern

landmasses to become recipients for glacial ice during the late

Pliocene. To complement the IP

25

record, we also studied the

distributions of glycerol dialkyl glycerol tetraethers (GDGTs) and

hydroxyl GDGTs to infer changes in SST

20

.

Results

Borehole chronology. The chronostratigraphic framework in

Hole 910C indicates the recovery of a complete Pliocene

sequence

21

with the age model revealing a consistently high

sedimentation rate (6–16 cm per kyr)

22

. The age model is

based on new paleomagnetic and biostratigraphic data of

Hole 910C and nearby Hole 911A (Fig. 1), and correlated into

high-resolution seismic data across the Yermak Plateau

22

(Supplementary Table 1). Additional fix points were derived

from correlating the global marine stack of d

18

O records

23

with

new benthic d

18

O data of Hole 910C (Supplementary Table 2).

The benthic d

18

O isotope record shows a gradual trend from

lower (B3.5–4.0%) to higher (B4.0–4.5%) values from the early

to the late Pliocene (Fig. 2), supporting the changes observed

elsewhere in the Nordic Seas, with similar or significantly less ice

volume occurring during the early Pliocene compared with the

late Pliocene

24

. The cooling in the late Pliocene can also be

inferred from the record of ice-rafted debris (IRD; Fig. 2), which

shows the largest IRD pulses concomitant with the intensification

of the Northern Hemisphere Glaciation (INHG) at

B2.64 Ma

ACEX 0° 70° 75° 80° 80° 75° 0° 910C 911A 10° GC15 PS57/166-2 EGC WSC TPD 910C

Figure 1 | Physiogeography of the Arctic Ocean. (a) Bathymetry and modern sea ice limit (white polygon) in September 2012 (National Snow and Ice Data Center (NSIDC)). Minimum (red line) and maximum (black line) sea ice extent for the interval 1981–2010 are also indicated (NSIDC). Locations of Ocean Drilling Program (ODP) Hole 910C and Integrated ODP Expedition 302 (ACEX—Arctic Coring Expedition) are indicated. (b) Surface currents in the Atlantic-Arctic gateway (TPD, Transpolar Drift; EGC, East Greenland Current; WSC, West Spitsbergen Current) and surface samples from two locations (PS57/166-2, 79°130_{N, 4°89}0_{E; GC15, 80°51}0_{N, 15°9}0_{E with IP}

25concentrations of 0.0086 and 0.0117 mg g 1sed., respectively. Locations of ODP Leg 151 Hole 910C and 911A are displayed.

(ref. 25), and occasional smaller IRD pulses at

B3.3 Ma (marine

isotope stage M2 glaciation)

26

and

B3.6 Ma (onset of Northern

Hemisphere Glaciation (NHG) sensu Mudelsee and Raymo

27

).

The latter is consistent with a distinct change from lower (3.6%)

to higher d

18

O values (4.4%) and falling sea level (  20 m; refs

28,29; Fig. 2), thus supporting inferences of enhanced ice volume

in the Northern Hemisphere since

B3.6 Ma (refs 27,30).

Pliocene sea ice record for the Arctic Ocean. Our IP

25

sea ice

reconstruction for the Fram Strait indicates ice-free conditions at

the borehole location between 5.8 and 3.9 Ma, as inferred from

the absence of IP

25

(Figs 2 and 3). This contrasts with the modern

setting, where IP

25

has been detected in surface sediments from

the region consistent with the site being located between the

winter and summer sea ice limits (Fig. 1)

18

. The failure to detect

IP

25

in sediments between 5.8 and 3.9 Ma may, potentially, also be

explained by the occurrence of perennial sea ice coverage.

However, this is unlikely since the interval is near the warm

period around the early/late Pliocene transition (B3.6 Ma)

recorded in the high Arctic

10

, with mean annual temperatures

significantly warmer than present

10,31

. The occurrence of the

phytoplanktic sterol brassicasterol (a proxy indicator of

open-water conditions

18

) (Fig. 3 and Supplementary Table 3) and

elevated SST estimates based on distributions of GDGTs and

hydroxyl GDGTs

32,33

(Fig. 3 and Supplementary Table 4) also

suggests an absence of perennial sea ice during the studied

interval. Distributions of the hydroxyl GDGTs, in particular,

indicate relatively warm conditions (ca. 7–10 °C and Fig. 3),

implying a dominance of Atlantic water rather than polar water

during the early Pliocene in the AAG region.

The first occurrence of IP

25

at ca. 3.9 Ma provides clear

evidence for the emergence of seasonal sea ice at the borehole

location, and thus a gradual expansion of Arctic sea ice cover

(Fig. 2). Following this onset, IP

25

concentrations remain

relatively low between ca. 3.9 and 3.0 Ma, consistent with less

sea ice coverage during the warm periods around the early/late

Pliocene transition and the mid-Piacenzian warm period from

2,500 3,000 3,500 4,000 4,500 5,000 Age (*1,000 years) Sea level (m) 4 0 5 10 15 20 25 0 0.002 0.004 0.006 0.008 0.01 0.012

Modern (winter) max.

Modern (summer) min.

M2 0 20 INHG _NHG 2,500 3,000 3,500 4,000 4,500 5,000 –80 –60 –40 –20 IRD (wt.% 100–1,000 μ m) Benthic δ 18 O (‰) 4.8 4.4 3.6 3.2 IP 25 ( μ g g –1 sed.)

Figure 2 | Pliocene proxy data from the AAG. (a) Pliocene sea level record29derived from global marine isotope stack23. (b) Stable oxygen isotopes (d18O) of benthic foraminifera Cassidulina teretis in Ocean Drilling Program (ODP) Hole 910C. Black line indicates the second-degree polynomial function. (c) Ice-rafted debris (IRD; wt.%, coarse fraction 100–1,000 mm) recorded from the eastern AAG (ODP Hole 911A; ref. 21). Black arrows indicate the intensification of the Northern Hemisphere Glaciation (INHG) atB2.64 Ma (ref. 25), as well as the onset of the Northern Hemisphere Glaciation (NHG) sensu Mudelsee and Raymo27. Marine Isotope Stage M2 glaciation atB3.30 Ma is also displayed. (d) IP25concentrations in ODP Hole 910C. Values representing proposed summer and winter sea ice limits in the study region are indicated by stippled lines.

3.29 to 2.97 Ma (refs 10,30,34,35) with sea ice conditions probably

similar to the modern (summer) minimum (Fig. 1). Nevertheless,

there is a gradual increase in IP

25

concentration up to ca. 2.7 Ma,

before a more rapid increase to modern maximum (winter) levels

(ca. 0.01 mg g

 1

sed.) (Fig. 2)

18

found in nearby surface sediments

(Fig. 1), suggesting a phased expansion towards contemporary sea

ice conditions at ca. 2.6 Ma, coincident with the INHG (that is,

glaciation to mid-latitudes) at ca. 2.64 Ma (ref. 25). A

semi-quantitative assessment of past sea ice conditions in the AAG

expressed by the PIP

25

index

18

suggests an extensive sea ice

cover after

B2.68 Ma as indicated by PIP

25

values 40.7

(Supplementary Fig. 1). IP

25

concentrations are also similar to

modern values over the last 2 Ma on the Yermak Plateau

19

, and

the new IP

25

record suggests the establishment of the modern

sea ice limits (Fig. 1) at ca. 2.6 Ma. Further, SST estimates based

on hydroxyl GDGTs (ca. 3.5–5.5 °C; Fig. 3 and Supplementary

Table 4) fall within the range of modern summer SSTs on the

southern Yermak Plateau (Fig. 4), indicating substantial

cooling and increasing influence of polar water masses parallel

to the onset of the NHG at

B3.6 Ma. Coldest SSTs (B3.5 °C)

after the INHG (Fig. 3) approach contemporary GDGT–SST

estimates observed in nearby surface sediments (Supplementary

Table 5).

Discussion

The implications of these observations are threefold: phase 1

(M/P boundary) is characterized by a dense vegetation cover in

the high northern latitudes, predominantly taiga forests with

dominant Picea and Pinus

36

, and a reduction of the Greenland

Ice Sheet by at least 30% (ref. 37). At this time, the Arctic Ocean

was partly isolated, with restricted AAG through-flow

21

, and the

Bering Strait still in the initial opening phase

38–41

. The prevalence

of an estuarine circulation in the Arctic Ocean

42,43

, together with

a high continental mean annual temperature (MAT) would have

restricted the occurrence of sea ice in the Arctic interior. As such,

the marginal seas (including the study location here) would have

been ice-free or covered by first-year winter sea ice only (Fig. 5).

During phase 2 (early/late Pliocene transition), changing

tectonic boundary conditions in the circum-Arctic during the

0 0.4 0.8 1.2 1.6 2 2.4 2 4 6 8 10 2,500 2,500 3,000 3,000 3,500 3,500 4,000 4,000 4,500 4,500 5,000 5,000 5,500 5,500 Age (*1,000 years) 0 0.002 0.004 0.006 0.008 0.01 0.012

Modern (summer) min.

INHG NHG

Modern (winter) max.

c

b

a

Brassicasterol ( μ g g –1 sed.) IP 25 (μ g g –1 sed.) SST (°C)

Figure 3 | Surface water conditions in the AAG during the Pliocene. (a) Sea surface temperature (SST) estimates. Values are based on distributions of GDGTs and hydroxyl GDGTs in Ocean Drilling Program (ODP) Hole 910C sediments (see Methods section for more details). Black arrows indicate the intensification of the Northern Hemisphere Glaciation (INHG) atB2.64 Ma (ref. 25), as well as the onset of the NHG sensu Mudelsee and Raymo27_{. (b) Brassicasterol (open-water indicator)}

concentrations. (c) IP25concentrations. Values representing proposed summer and winter sea ice limits in the study region are indicated by stippled lines. –5 0 5 10 4,000 3,000 2,000 1,000 0

Ocean data view

Temperature (°C) Depth (m) –4 –2 0 2 4 82°N 83°N 84°N 85°N

0° _{10°E 20°E} 30°E

0°

10°E 20°E 30°E

Ocean data view

82°N 83°N

84°N 85°N

Ocean data view

WOD 2013, stations Temperature (°C), depth = 50 m 81°N 80°N 81°N 80°N

Figure 4 | Modern Oceanography. (a) Available stations in the study region stored in World Ocean Database (WOD) 2013. (b) Sea surface temperature (depth¼ 50 m) distribution in the study region. Black circle indicates the position of ODP Hole 910C. (c) Temperature-Depth profile of all stations in the study region. Red line indicates the reference station at the borehole location (WOD13_DE) measured at 19 June 1984.

late Miocene-early Pliocene, including mountainous uplift

44

and

open/closing gateways (AAG, Bering Strait, Central American

Seaways (CAS))

21,45,46

, triggered the expansion of the Arctic sea

ice coverage, reaching the modern summer limit (Fig. 1) for the

first time at ca. 3.9 Ma (Fig. 5). This timing is consistent with

recent inferences on tectonic uplift events on Greenland

47

, the

Barents Sea

21

and Arctic Canada

48

, when elevated plateaus

became available to host perennial ice fields and, ultimately,

continental ice. The expansion of sea ice in the Arctic at this time

parallels the increased low-saline Arctic through-flow via Bering

Strait as documented by a major invasion of Pacific mollusks to

the North Atlantic from

B4.4 to 4.5 Ma (ref. 49). The evidence of

an oscillating West Antarctic ice sheet

50

and sea level variations on

average by ±20 m around the modern level until 3.3 Ma

(refs 28,29; Fig. 2) suggest that, between 5.0 and 3.9 Ma, the

Bering Strait was mostly submerged. Enhanced freshwater delivery

to the Arctic via Siberian rivers associated with the advancing

closure of the CAS and increased MOC between 4.7 and 4.5 Ma

(refs 46,51,52) further facilitated sea ice formation in the Arctic

Ocean and supports the inferred establishment of the

‘modern-type’ East Greenland Current between 4.5 and 4.3 Ma (ref. 21).

The enhanced sea ice export through the (now) established

deep-water AAG counterbalanced the heat transport associated

with the enhanced MOC owing to the closure of the CAS.

In contrast, modern sea ice limits, with the margin located

above the borehole location throughout most of the year, were

not reached until the INHG at ca. 2.64 Ma (phase 3) (Fig. 5). The

unprecedented long period of ice marginal conditions in the AAG

parallel to the INHG confirms the amplification of polar cooling

in response to internal feedbacks superimposed on the long-term

decrease in atmospheric CO

2

concentration and slow tectonic

forcings

27,53

. Among the array of feedbacks that are debated, the

most sensitive for sea ice and glacial ice conditions in the AAG

region are likely changes in the MOC in response to the final

closure of the CAS after ca. 2.9 Ma (ref. 54). That the closure and

associated enhanced MOC is unable to produce large-scale ice

sheets in numerical models

55

may be owing to the use of similar

to modern boundary conditions (gateways and orography) for the

Pliocene. However, adjusting the models for orographic changes

over Greenland, and the exposed landmasses of northern Eurasia

above a threshold for glacial ice accumulation and massive winter

sea ice coverage since ca. 4 Ma (Fig. 5), may allow models to

simulate enhanced MOC and more precipitation over Greenland

and reproduce the formation of large-scale ice sheets. The

additional effect of the irreversible closure of the CAS after ca.

2.9 Ma (that is, increased poleward atmospheric moisture

transport—both in the AAG and the subarctic Pacific

56

)

induced a freshening of the Arctic Ocean

54

, as evidenced by the

rapid spatial and temporal expansion of its sea ice cover. These, in

turn, strengthened the sea ice export via the East Greenland

Current,

promoted

ice-albedo

feedback

mechanisms

and

thermally isolated Greenland; thus, fostering build-up of

continental ice sheets in the circum-Arctic during the INHG.

The pre-requisite for these marked changes in the polar

ocean-ice-continent climate system, however, were orographic changes

in the circum-Arctic and freshening of the Arctic Ocean during

the early Pliocene. Pre-glacial uplift and final deepening of the

gateways preconditioned the landmasses to become recipient for

glacial ice. Before the closure of the CAS as an additional

amplifier for Arctic freshening and sea ice expansion

57

, we

suggest that tectonic forcing in the Arctic was the decisive factor

responsible

for

climate

deterioration

in

the

Northern

Hemisphere. Finally, the new sea ice reconstruction for the

Pliocene Arctic Ocean provides a new dimension for paleoclimate

modelling to reproduce a warm climate state in Earth history

such as the mid Pliocene warmth

34

. Thus, by demonstrating that

sea ice in the Eurasian sector of the Arctic existed along its

modern summer limit during the most recent interval of

long-term average warmth relative to the last million years, climate

simulations, with fully proxy-consistent boundary conditions for

the Arctic Ocean, can be further refined to improve validation of

projected climate for the end of the current century

1

.

Methods

Stable isotope analyses

.

The new stable isotope data set of the benthic for-aminifera Cassidulina teretis test (100–1,000 mm fraction) was generated at the

910

ACEX

BS

HR +

Phase 1 (Miocene/Pliocene transition) Phase 2 (early/late Pliocene, ~4 Ma) Phase 3 (late Pliocene, ~2.6 Ma)

ACEX + + + 910 ACEX + + 910 ACEX BS + + + + Gr SBa Ma Al ES BS

Figure 5 | Development of modern sea ice cover during the Pliocene. (Phase 1) Perennial sea ice in the Arctic interior during the Miocene/Pliocene transition. Late Miocene/early Pliocene tectonic uplift in circum-Arctic (crosses) is recorded in the Svalbard/Barents Sea region (SBa), on Greenland (Gr), Mackenzie Region (Ma), Alaska (Al) and East Siberia (ES)21,44,47,48. Early Pliocene vegetation is characterized by taiga forests with dominant Picea and Pinus36. Restricted deep/shallow water exchange through Atlantic-Arctic gateway (HR, Hovgaard Ridge)21and Bering Strait (BS)38–41. (Phase 2) Arctic sea ice expanded to its modern summer limits for the first time afterB4 Ma. Pre-glacial uplift reached a threshold for ice-growth (stippled lines). Enhanced thermohaline circulation (red/blue arrows in AAG)46,51and freshening of the Arctic through BS and Siberian rivers51,57(black arrows). (Phase 3) Expansion of Arctic sea ice to its modern winter limits atB2.6 Ma. Stippled blue line indicates the modern winter sea ice limits in the AAG. Large-scale ice sheets recorded in Eurasia21_{, Greenland}54_{and North America}70_{close to potential moisture sources in the Nordic seas (red open circle)}54_{and subarctic Pacific}

Leibniz Laboratory for Radiometric Dating and Stable Isotope Research in Kiel, Germany, using a Finnigan MAT 251 isotope ratio gas mass spectrometer, directly coupled to an automated carbonate preparation device (Kiel I prototype) and calibrated using NIST19 international standard to the Vienna Peedee belemnite scale. The precision of the measurements at 1s based on repeated analyses of internal laboratory standards was better than ±0.07% and ±0.05% for oxygen and carbon isotopes, respectively. Because C. teretis secretes its carbonate close to oxygen isotopic equilibrium with the ambient seawater, no further corrections or adjustments were applied to the d18O data58.

Biomarker analyses

.

IP25was extracted, identified and quantified according to the methods of Belt et al.59_{with minor modifications. Briefly, 9-octylheptadec-8-ene} (10 ml; 10 mg ml 1) was added as an internal standard to each freeze-dried sediment sample (2–3 g) to permit subsequent quantification by gas

chromatography—mass spectrometry (GC–MS). Sediments were then extracted using a mixture of dichloromethane/methanol (2:1, v/v) and ultrasonication (15 min), and the resulting solutions were decanted and dried (N2) to yield a total organic extract (TOE). Before analysis of TOEs using GC–MS, it was necessary to first remove elemental sulphur that was present in relatively high concentrations in the majority of the extracts. This was achieved by first re-dissolving the TOEs in hexane (1 ml) and adding tetrabutylammonium sulphite (1 ml) and 2-propanol (2 ml). The resulting suspensions were then agitated by hand (1 min) before addition of a further 3 ml of water and re-agitated (1 min). Samples were then centrifuged (2 min) and the surface (hexane) layer containing IP25was transferred to a clean vial. The aqueous phase was extracted twice more (hexane; 2  1 ml). Following removal of hexane from the combined extracts (nitrogen; room temperature), nonpolar lipids (including IP25) were obtained using open-column chromatography (silica; 6 ml hexane). These partially purified nonpolar lipid fractions were further separated into saturated and unsaturated hydrocarbons using glass pipettes containing silver ion solid phase extraction material (Supelco discovery Ag-ion). Saturated hydrocarbons (hexane; 5 ml, then dichloromethane; 5 ml) and unsaturated hydrocarbons including IP25(dichloromethane/acetone (95/5); 10 ml) were eluted as consecutive fractions. Individual hydrocarbon fractions were analysed by GC–MS with operating conditions as described previously, for example, ref. 59. Mass spectrometric analysis was carried out in total ion current and single-ion monitoring (SIM) modes. IP25was identified on the basis of its characteristic GC retention index and mass spectrum obtained from a laboratory standard (Supplementary Fig. 2). The identification of IP25was further confirmed by hydrogenation (PtO2.2H2O; H2; 30 min) to the parent C25highly branched isoprenoid alkane, which was identified on the basis of its own characteristic retention index and mass spectrum and by comparison of each of these with those obtained from a laboratory standard. Quantification was achieved by dividing the integrated GC–MS peak area of IP25by that of the internal standard (9-octylheptadec-8-ene; both m/z (mass to charge ratio) 350) and normalizing this ratio using an instrumental response factor obtained from laboratory standards of each analyte59_{and the mass of sediment. The analytical reproducibility (7%, n ¼ 3)} was determined using a standard sediment with a known concentration of IP25. The limit of detection (s/n ¼ 3) for IP25was 0.5 ng g 1dry sediment based on extraction of 2 g sediment and concentration of partially purified sediment extracts to 20 ml before analysis by GC–MS. Analysis of brassicasterol was carried out as per Belt et al.60Calculation of PIP25values was performed according to Mu¨ller et al.18

For the GDGTs analyses, freeze-dried sediment samples (3–6.7 g) were extracted with a solvent mixture of dichloromethane/methanol (3:1,v/v) using microwave assisted extraction in a MARS X System (CEM, Matthews, NC, USA). After centrifugation and evaporation of the solvent, the extracts were separated into four fractions using 1% deactivated SiO2column chromatography. The eluents were hexane, hexane/dichloromethane (2:1, v/v), dichloromethane and dichloromethane/methanol (95:5, v/v). GDGTs were obtained in the fourth fraction, dried under gentle N2-stream, re-dissolved in hexane/n-propanol (99:1, v/ v) and filtered through 0.50 mm polytetrafluoroethylene (PTFE) filters (Advantec). Extracts were then eluted through a Tracer Excel CN column (0.4  20 cm, 3 mm; Teknokroma), equipped with a pre-column filter and a guard column using high-performance liquid chromatography coupled to mass spectrometry (HPLC-MS). using an atmospheric pressure chemical interphase. The solvent programme was adapted from Schouten et al.61_{and Escala et al.}62_{Samples were eluted with hexane/} n-propanol at 0.6 ml min 1. The amount of n-propanol was held at 1.5% for 4 min, increased gradually to 5.0% during 11 min, then increased to 10% during 1 min and held at 10% for 4 min, then decreased to 1.5% during 1 min and held at 1.5% for 9 min until the end of the run. The parameters of the atmospheric pressure chemical interphase were set as follows to generate positive ion spectra: corona discharge 3 mA, vaporizer temperature 400 °C, sheath gas pressure 49 mTorr, auxiliary gas (N2) pressure 5 mTorr and capillary temperature 200 °C. Isoprenoid GDGTs were monitored in selected ion monitoring (SIM) mode at m/z 1,302, 1,300, 1,298, 1,296 and 1,292. The hydroxyl GDGTs were monitored in SIM mode at m/z 1,318 and 1,316. The identification of the hydroxyl GDGTs at m/z 1,318 and 1,316 has been described in Huguet et al.33The HPLC–MS system was checked for TEX86index accuracy with a standard sediment sample before and between sample sets63_{. The synthetic tetraether lipid GR was used as external} standard. Compound GR has an m/z of 1,208, a structure typical of neutral archaeal membrane lipids and presumably does not occur in the environment63,64.

External curves were measured before each sample series. The reproducibility of the quantification is estimated to be ±10%.

When temperatures are expected to fall below 15 °C, Kim et al.65recommended the use of the TEX86L index to estimate past SST (error estimates (1 sigma) are given as 4 °C (ref. 65)): TEX86L¼ log[GDGT2/(GDGT1 þ GDGT2 þ GDGT3)]; SST ¼ 67.5  TEX86L þ 46.9 (r2¼ 0.86, n ¼ 396, Po0.0001). It should be noted, however, that SST estimates derived from both TEX86or TEX86L may be anomalously high in the Arctic, especially in the vicinity of Siberian river mouths and the sea ice margin66. A modified TEX86index and calibration (TEX860) had been proposed and applied for Arctic Ocean (B87.5° N) temperatures during the Palaeocene/Eocene thermal maximum67_:

TEX860¼ [(GDGT2 þ GDGT3 þ Cren0)/(GDGT1 þ GDGT2 þ Cren0)]; TEX860¼ 0.016  SST þ 0.20 (R2¼ 0.93). TEX860, however, also yields anomalously high temperatures in Hole 910C (Supplementary Table 4).

The recently discovered hydroxyl GDGTs68_{have been shown to infer SSTs on a} global scale33and to indicate the presence of polar waters in the Arctic Ocean32. Fietz et al.32_{proposed a so-called OH-GDGT% index and a promising SST} calibration (slightly modified from Huguet et al.33_{) based on the hydroxyl and} isoprenoid GDGTs:

OH-GDGT% ¼ (S hydroxyl GDGTs)/(S hydroxyl GDGTs þ S isoprenoid GDGTs); OH-GDGT% ¼ 8.6–0.67  SST (R2_{¼ 0.55, P ¼ 0.004, n ¼ 11). The} OH-GDGT% index yields reasonable SSTs for Hole 910C (Supplementary Tables 4 and 5) even though absolute values must be confirmed in larger surface sediment data sets.

Modern oceanography

.

Modern SST (depth ¼ 50 m) data from the study region were extracted from the World Ocean Database (WOD 2013) and displayed by ODV version 4.6.2 (ref. 69).

References

1. IPCC. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Changes (eds Stocker, T. F. et al.) 1535 (Cambridge Univ. Press, 2013).

2. Backman, J., Moran, K. & McInroy, D., and the IODP Expedition 302 Scientists. IODP Expedition 302, Arctic Coring Expedition (ACEX): a first look at the Cenozoic paleoceanography of the central Arctic Ocean. Sci. Drill. 1, 12–17 (2005).

3. Polyak, L. et al. History of sea ice in the Arctic. Quat. Sci. Rev. 29, 1757–1778 (2010).

4. Darby, D. A. Arctic perennial ice cover over the last 14 million years. Paleoceanography 23, 1–9 (2008).

5. Krylov, A. A. et al. A shift in heavy and clay mineral provenance indicates a middle Miocene onset of a perennial sea ice cover in the Arctic Ocean. Paleoceanography 23, 1–10 (2008).

6. Darby, D. A. Ephemeral formation of perennial sea ice in the Arctic Ocean during the middle Eocene. Nat. Geosci. 7, 210–213 (2014).

7. Matthiessen, J., Brinkhuis, H., Poulsen, N. & Smelror, M. Decahedrella martinheadii Manum 1997-a stratigraphically and paleoenvironmentally useful Miocene acritarch of the high northern latitudes. Micropaleontology 55, 171–186 (2009).

8. Cronin, T. M. et al. Microfaunal evidence for the elevated Pliocene temperatures in the Arctic Ocean. Paleoceanography 8, 161–173 (1993). 9. Ballantyne, A. P. et al. The amplification of Arctic terrestrial surface

temperatures by reduced sea-ice extent during the Pliocene. Palaeogeogr. Palaeoclimatol. Palaeoecol. 386, 59–67 (2013).

10. Rybczynski, N. et al. Mid-Pliocene warm-period deposits in the High Arctic yield insight into camel evolution. Nat. Commun. 4, 1550 (2013). 11. Haywood, A. M. & Valdes, P. J. Modelling Pliocene warmth: contribution of

atmosphere, oceans and cryosphere. Earth Planet. Sci. Lett. 218, 363–377 (2004). 12. Dowsett, H. J. et al. Sea surface temperature of the mid-Piacenzian Ocean: a

data-model comparison. Sci. Rep. 3, 2013 (2013).

13. Myhre, A. M. et al. Proceedings Ocean Drilling Program. Initial Reports, Vol. 151, 926 (Texas A&M University, 1995).

14. Belt, S. T. et al. A novel chemical fossil of palaeo sea ice: IP25. Org. Geochem. 38, 16–27 (2007).

15. Brown, T. A., Belt, S. T., Tatarek, A. & Mundy, C. J. Source identification of the Arctic sea ice proxy IP25. Nat. Commun. 5, 4197 (2014).

16. Belt, S. T. & Mu¨ller, J. The Arctic sea ice biomarker IP25: a review of current understanding, recommendations for future research and applications in palaeo sea ice reconstructions. Quat. Sci. Rev. 79, 9–25 (2013).

17. Mu¨ller, J., Masse, G., Stein, R. & Belt, S. T. Variability of sea-ice conditions in the Fram Strait over the past 30.000 years. Nat. Geosci. 2, 772–776 (2009). 18. Mu¨ller, J. et al. Towards quantitative sea ice reconstructions in the northern

North Atlantic: A combined biomarker and numerical modelling approach. Earth Planet. Sci. Lett. 306, 137–148 (2011).

19. Stein, R. & Fahl, K. Biomarker proxy shows potential for studying the entire Quaternary Arctic sea ice history. Org. Geochem. 55, 98–102 (2013).

20. Schouten, S., Hopmans, E. C. & Sinninghe Damste´, J. S. The organic geochemistry of glycerol dialkyl glycerol tetraether lipids: a review. Org. Geochem. 54, 19–61 (2013).

21. Knies, J. et al. Effect of early Pliocene uplift on late Pliocene cooling in the Arctic-Atlantic gateway. Earth Planet. Sci. Lett. 387, 132–144 (2014). 22. Mattingsdal, R. et al. A new 6 Myr stratigraphic framework for the

Atlantic-Arctic Gateway. Quat. Sci. Rev. 92, 170–178 (2014).

23. Lisiecki, L. E. & Raymo, M. E. A Pliocene-Pleistocene stack of 57 globally distributed benthic delta O-18 records. Paleoceanography 20, 1–17 (2005). 24. Fronval, T. & Jansen, E. in Proceedings Ocean Drilling Program Vol. 151

(eds Thiede, J., Myhre, A. M., Firth, J. V., Johnson, G. L. & Ruddiman, W. F.) 455–468 (Texas A&M University, 1996).

25. Bailey, I. et al. An alternative suggestion for the Pliocene onset of major northern hemisphere glaciation based on the geochemical provenance of North Atlantic Ocean ice-rafted debris. Quat. Sci. Rev. 75, 181–194 (2013). 26. De Schepper, S. et al. Northern Hemisphere Glaciation during the globally

warm Early Late Pliocene. PLoS ONE 8, e81508 (2013).

27. Mudelsee, M. & Raymo, M. E. Slow dynamics of the Northern Hemisphere glaciation. Paleoceanography 20, 1–14 (2005).

28. Miller, K. G. et al. The phanerozoic record of global sea-level change. Science 310,1293–1298 (2005).

29. Miller, K. G., Mountain, G. S., Wright, J. D. & Browning, J. V. A 180 million year record of sea level and ice volume variations from continental margin and deep sea isotopic records. Oceanography 24, 40–53 (2011).

30. De Schepper, S., Gibbard, P. L., Salzmann, U. & Ehlers, J. A global synthesis of the marine and terrestrial evidence for glaciation during the Pliocene Epoch. Earth Sci. Rev. 135, 83–102 (2014).

31. Ballantyne, A. P. et al. Significantly warmer Arctic surface temperatures during the Pliocene indicated by multiple independent proxies. Geology 38, 603–606 (2010).

32. Fietz, S., Huguet, C., Rueda, G., Hambach, B. & Rosell-Mele, A. Hydroxylated isoprenoidal GDGTs in the Nordic Seas. Marine Chem. 152, 1–10 (2013). 33. Huguet, C., Fietz, S. & Rosell-Mele, A. Global distribution patterns of hydroxy

glycerol dialkyl glycerol tetraethers. Org. Geochem. 57, 107–118 (2013). 34. Dowsett, H. J. et al. Assessing confidence in Pliocene sea surface temperatures

to evaluate predictive models. Nat. Clim. Change 2, 365–371 (2012). 35. Prescott, C. L. et al. Assessing orbitally-forced interglacial climate variability

during the mid-Pliocene Warm Period. Earth Planet. Sci. Lett. 400, 261–271 (2014).

36. Salzmann, U. et al. Climate and environment of a Pliocene warm world. Palaeogeogr. Palaeoclimatol. Palaeoecol. 309, 1–8 (2011).

37. Dolan, A. M. et al. Sensitivity of Pliocene ice sheets to orbital forcing. Palaeogeogr. Palaeoclimatol. Palaeoecol. 309, 98–110 (2011).

38. Gladenkov, A. Y., Oleinik, A. E., Marincovich, L. & Barinov, K. B. A refined age for the earliest opening of Bering Strait. Palaeogeogr. Palaeoclimatol. Palaeoecol. 183,321–328 (2002).

39. Gladenkov, A. Y. & Gladenkov, Y. B. Onset of connections between the Pacific and Arctic Oceans through the Bering Strait in the Neogene. Stratigr. Geol. Correl. 12, 175–187 (2004).

40. Marincovich, L. & Gladenkov, A. Y. New evidence for the age of Bering Strait. Quat. Sci. Rev. 20, 329–335 (2001).

41. Marincovich, L. & Gladenkov, A. Y. Evidence for an early opening of the Bering Strait. Nature 397, 149–151 (1999).

42. Jakobsson, M. et al. The early Miocene onset of a ventilated circulation regime in the Arctic Ocean. Nature 447, 986–990 (2007).

43. Matthiessen, J., Knies, J., Vogt, C. & Stein, R. Pliocene palaeoceanography of the Arctic Ocean and subarctic seas. Phil. Trans. R. Soc. A 367, 21–48 (2009). 44. Green, P. F. & Duddy, I. R. Synchronous exhumation events around the Arctic

including examples from Barents Sea and Alaska North Slope. Petrol. Geol. Conf. Ser. 7, 633–644 (2010).

45. Marincovich, L. Central American paleogeography controlled Pliocene Arctic Ocean molluscan migrations. Geology 28, 551–554 (2000).

46. Haug, G. H. & Tiedemann, R. Effect of the formation of the Isthmus of Panama on Atlantic Ocean thermohaline circulation. Nature 393, 673–676 (1998). 47. Japsen, P., Green, P. F., Bonow, J. M., Nielsen, T. F. D. & Chalmers, J. A. From

volcanic plains to glaciated peaks: Burial, uplift and exhumation history of southern East Greenland after opening of the NE Atlantic. Global Planet. Change 116, 91–114 (2014).

48. McNeil, D. H. et al. Sequence stratigraphy, biotic change, Sr-87/Sr-86 record, paleoclimatic history, and sedimentation rate change across a regional late Cenozoic unconformity in Arctic Canada. Can. J. Earth Sci. 38, 309–331 (2001). 49. Verhoeven, K., Louwye, S., Eiriksson, J. & De Schepper, S. A new age model for

the Pliocene-Pleistocene Tjo¨rnes section on Iceland: Its implication for the timing of North Atlantic-Pacific paleoceanographic pathways. Palaeogeogr. Palaeoclimatol. Palaeoecol. 309, 33–52 (2011).

50. Naish, T. et al. Obliquity-paced Pliocene West Antarctic ice sheet oscillations. Nature 458, 322–U384 (2009).

51. Haug, G. H., Tiedemann, R., Zahn, R. & Ravelo, A. C. Role of Panama uplift on oceanic freshwater balance. Geology 29, 207–210 (2001).

52. Driscoll, N. W. & Haug, G. H. A short circuit in thermohaline circulation: A cause for northern hemisphere glaciation? Science 282, 436–438 (1998). 53. Cane, M. A. & Molnar, P. Closing of the Indonesian seaway as a precursor to east

African aridircation around 3-4 million years ago. Nature 411, 157–162 (2001). 54. Sarnthein, M. et al. Mid-Pliocene shifts in ocean overturning circulation and

the onset of Quaternary-style climates. Clim. Past 5, 269–283 (2009). 55. Lunt, D. J., Valdes, P. J., Haywood, A. & Rutt, I. C. Closure of the Panama

Seaway during the Pliocene: implications for climate and Northern Hemisphere glaciation. Clim. Dyn. 30, 1–18 (2008).

56. Haug, G. H. et al. North Pacific seasonality and the glaciation of North America 2.7 million years ago. Nature 433, 821–825 (2005).

57. Driscoll, N. W. & Haug, G. H. A short circuit in thermohaline circulation: A cause for Northern Hemisphere Glaciation. Science 282, 436–438 (1998). 58. Jansen, E., Bleil, U., Henrich, R., Kringstad, L. & Slettemark, B.

Paleoenvironmental changes in the Norwegian sea and the northeast Atlantic during the last 2.8 m.y.: Deep Sea Drilling Project/Ocean Drilling Program Sites 610, 642, 643 and 644. Paleoceanography 3, 563–581 (1988).

59. Belt, S. T. et al. A reproducible method for the extraction, identification and quantification of the Arctic sea ice proxy IP25from marine sediments. Anal. Methods 4, 705–713 (2012).

60. Belt, S. T. et al. Quantitative measurement of the sea ice diatom biomarker IP25 and sterols in Arctic sea ice and underlying sediments: Further considerations for palaeo sea ice reconstruction. Org. Geochem. 62, 33–45 (2013).

61. Schouten, S., Huguet, C., Hopmans, E. C., Kienhuis, M. V. M. & Damste, J. S. S. Analytical methodology for TEX86paleothermometry by high-performance liquid chromatography/atmospheric pressure chemical ionization-mass spectrometry. Anal. Chem. 79, 2940–2944 (2007).

62. Escala, M., Rosell-Mele, A. & Masque, P. Rapid screening of glycerol dialkyl glycerol tetraethers in continental Eurasia samples using HPLC/APCI-ion trap mass spectrometry. Org. Geochem. 38, 161–164 (2007).

63. Escala, M. Application of Tetraether Membrane Lipids as Proxies for Continental Climate Reconstruction in Iberian and Siberian lakes. PhD thesis, Universitat Autonoma de Barcelona (2009).

64. Rethore, G. et al. Archaeosomes based on synthetic tetraether-like lipids as novel versatile gene delivery systems. Chem. Commun. 28, 2054–2056 (2007). 65. Kim, J. H. et al. New indices and calibrations derived from the distribution of

crenarchaeal isoprenoid tetraether lipids: Implications for past sea surface temperature reconstructions. Geochim. Cosmochim. Acta 74, 4639–4654 (2010). 66. Ho, S. L. et al. Appraisal of TEX86 and TEX86L thermometries in subpolar and

polar regions. Geochim. Cosmochim. Acta 131, 213–226 (2014).

67. Sluijs, A. et al. Subtropical arctic ocean temperatures during the Palaeocene/ Eocene thermal maximum. Nature 441, 610–613 (2006).

68. Liu, X.-L. et al. Mono- and dihydroxyl glycerol dibiphytanyl glycerol tetraethers in marine sediments: Identification of both core and intact polar lipid forms. Geochim. Cosmochim. Acta 89, 102–115 (2012).

69. Schlitzer, R. Ocean Data View http://odv.awi.de (2012).

70. Balco, G., Rovey, C. W. & Stone, J. O. H. The first glacial maximum in North America. Science 307, 222–222 (2005).

Acknowledgements

This research used samples and data provided by the Integrated Ocean Drilling Program and the British Ocean Sediment Core Research Facility. This research is part of the Centre of Excellence ‘CAGE—Arctic Gas hydrate, Environment and Climate’ (Norwe-gian Research Council (NRC) grant 223259) at the University of Tromsø, Norway and was funded by Statoil ASA, Det Norske Oljeselskap and BG Group, as well as the NRC grant 200672/S60. We acknowledge Lukas Smik (Plymouth) for providing us with sedimentary IP25data from surface sediment material (GC15) and to Michael Wilde

(Plymouth) for assistance in obtaining mass spectra. We thank Gerald H. Haug and Jens Matthiessen for stimulating discussion. Laura Falco´ and Gemma Rueda are acknowl-edged for the analysis of, and discussion on, GDGTs.

Author contributions

The main idea was developed by J.K. and J.K. and S.B.T. wrote most of the text; the IP25

analyses were made by P.C.-S, while GDGT data were obtained by S.F. and A.R.-M; S.B. produced the stable isotope record; and all authors discussed the results and commented on the manuscript.

Additional information

Supplementary Informationaccompanies this paper at http://www.nature.com/ naturecommunications

Competing financial interests:The authors declare no competing financial interests. Reprints and permissioninformation is available online at http://npg.nature.com/ reprintsandpermissions/

How to cite this article:Knies, J. et al. The emergence of modern sea ice cover in the Arctic Ocean. Nat. Commun. 5:5608 doi: 10.1038/ncomms6608 (2014).